Graft-transmissible induction of potato tuberization by the microRNA miR172.

The photoreceptor phytochrome B (PHYB) and the homeodomain protein BEL5 are involved in the response of potato tuber induction to the photoperiod. However, whether they act in the same tuberization pathway is unknown. Here we show the effect of a microRNA, miR172, on this developmental event. miR172 levels are higher under tuber-inducing short days than under non-inductive long days and are upregulated in stolons at the onset of tuberization. Overexpression of this microRNA in potato promotes flowering, accelerates tuberization under moderately inductive photoperiods and triggers tuber formation under long days. In plants with a reduced abundance of PHYB, which tuberize under long days, both BEL5 mRNA and miR172 levels are reduced in leaves and increased in stolons. This, together with the presence of miR172 in vascular bundles and the graft transmissibility of its effect on tuberization, indicates that either miR172 might be mobile or it regulates long-distance signals to induce tuberization. Consistent with this, plants overexpressing miR172 show increased levels of BEL5 mRNA, which has been reported to be transmissible through grafts. Furthermore, we identify an APETALA2-like mRNA containing a miR172 binding site, which is downregulated in plants overexpressing miR172 and plants in which PHYB is silenced. Altogether, our results suggest that miR172 probably acts downstream of the tuberization repressor PHYB and upstream of the tuberization promoter BEL5 and allow us to propose a model for the control of tuberization by PHYB, miR172 and BEL5.


INTRODUCTION
Like flowering in many species, potato (Solanum tuberosum L.) tuberization is regulated by day length. In the S. tuberosum subspecies andigena, short days (SDs) strongly induce tuberization, short days supplemented with a night break (SD+NB) are moderately inductive and long days (LDs) are repressive (Rodríguez-Falcón et al., 2006;González-Schain and Suárez-López, 2008). Very few genes involved in the photoperiodic control of tuber induction have been identified and their possible functions in the timing of tuberization have hardly been explored.
The photoreceptor phytochrome B (PHYB) plays an important role in the response of tuberization to day length. PHYB is required to inhibit tuberization under LD conditions, as PHYB antisense (anti-PHYB) plants tuberize under LDs (Jackson et al., 1996). This effect is transmitted through grafts, which has been interpreted as evidence that PHYB induces the generation of a mobile inhibitor of tuberization (Jackson et al., 1998) that must be produced in the leaves in response to LDs. However, the mechanism by which PHYB controls photoperiodic tuberization is unknown. Although it has been shown that PHYB affects the mRNA levels of a gibberellin-20-oxidase (GA20ox1), a key enzyme in gibberellin biosynthesis that affects tuberization, the alteration of GA20ox1 transcript and GA 1 levels in anti-PHYB plants is not consistent with their tuberization phenotype Jackson et al., 2000;Martínez-García et al., 2002). Therefore, how PHYB negatively regulates tuberization has not been elucidated.
A homeodomain protein, BEL5, and a sucrose transporter, SUT4, are also involved in the control of tuberization by the photoperiod (Chen et al., 2003;Banerjee et al., 2006b;Chincinska et al., 2008). BEL5 overexpression promotes tuberization and overrides the inhibition caused by LDs (Chen et al., 2003;Banerjee et al., 2006b). A recent report suggests that BEL5 mRNA is graft transmissible and its movement to stolons correlates with tuber induction (Banerjee et al., 2006b). By contrast, SUT4 is a tuberization inhibitor, as SUT4 RNAi plants tuberize under non-inductive LDs (Chincinska et al., 2008). In addition, SUT4 RNAi plants flower earlier than the wild type (Chincinska et al., 2008), revealing another link in the regulation of flowering and tuber induction.
The miR172 sequence had been previously found in a potato EST (Aukerman and Sakai, 2003), suggesting that this miRNA is produced in potato. Given the similarities in the control of flowering and tuberization (Suárez-López, 2005; Rodríguez-Falcón et al., 2006), we tested the hypothesis that miR172 can regulate flowering and tuber induction in this plant species. Our results show that miR172 affects the induction of tuberization, probably upstream of BEL5 and, like in Arabidopsis, is involved in the temporal control of flowering. The study of miR172 abundance and location strongly suggests that miR172 acts downstream of PHYB to regulate tuber induction and that this miRNA is a component or a regulator of longdistance signals. Furthermore, the results suggest that PHYB acts upstream of BEL5. The cloning and expression analysis of a potato AP2-like gene harboring a miR172 target site indicates that this miRNA negatively regulates its target transcripts in potato. Finally, we propose a model for a pathway regulating potato tuberization.

Plant material and growth conditions
Solanum tuberosum L. subspecies andigena line 7540 was used as wild-type potato. Anti-PHYB plants were kindly provided by Salomé Prat (Jackson et al., 1996). Plants were propagated from single-node stem cuttings on Murashige and Skoog (MS) medium containing 20 g l -1 sucrose and 2 g l -1 Gelrite (Duchefa Biochemie, Haarlem, The Netherlands) at 23°C under LD conditions (16 hours light and 8 hours darkness). Two-week-old plants were planted in soil composed of substrate type 3 (Plantaflor) and sand (3:1) and grown in the greenhouse at 23°C under LD conditions. During autumn and winter, the light period was extended to 16 hours with high-pressure sodium vapor lamps (SON-T Agro 400 W, Philips Ibérica, Madrid, Spain). Whenever natural light intensity fell below 65 μmol m -2 second -1 during the light period, light intensity was supplemented with the same lamps. Plants were watered daily with modified Hoagland's solution (Johnson et al., 1957) diluted 1/60.

Prediction of miR172 precursor secondary structure
The first 500 nucleotides of the reverse complement of a putative potato miR172 precursor (accession number BQ114970) were used to predict its secondary structure using Mfold (http://www.bioinfo.rpi.edu/applications/ mfold) (Zuker, 2003). Nine different structures were predicted, with ΔG values ranging from -110.77 to -99.93 kcal mol -1 . The structure with the lowest ΔG value was chosen as representative and is shown in Fig. S1 in the supplementary material.

Analysis of miR172 levels
Total RNA was extracted using TRIzol Reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions with the following modification: the isopropanol precipitation was done with a high-salt solution (1.2 M NaCl and 0.8 M sodium citrate) to remove polysaccharide contaminations. Alternatively, total RNA was isolated using the Real ARNzol Spin Kit (+PVP) (Durviz, Valencia, Spain).

In situ hybridization
Plant material was fixed for 8 hours at 4°C in 4% paraformaldehyde and 0.1 M phosphate buffer pH 7, dehydrated through a graded series of ethanol and butanol dilutions, embedded in Paraplast Plus (Paraplast X-Tra, Oxford Labware, St Louis, MO, USA) and sectioned to 7 μm with a microtome. Tissue sections were deparaffinized with Histoclear, rehydrated through an ethanol series, and then pre-treated with proteinase K (0.1 U ml -1 ) in Tris-HCl pH 7.5 at 37°C for 15 minutes. Digestion was stopped by washing with 0.2ϫ PBS containing 0.04 M glycine and then twice with PBS for 2 minutes. After dehydrating in ethanol baths, hybridization was performed at 37°C overnight with 0.02 μM digoxigenin-labeled miRCURY LNA probe complementary to miR172 (Exiqon, Vedback, Denmark) in hybridization solution (50% formamide, 2ϫ SSC, 4ϫ Denhardt's solution, 20% dextran, 2 mg ml -1 tRNA). An LNA oligonucleotide complementary to mouse miR124 (5Ј-TTAAGGCACGCGGTGAATGCCA-3Ј), with no predicted target sequences in plants, was used as negative control. After hybridization, slides were washed in 2ϫ SSC at 37°C for 45 minutes and in 1ϫ SSC for 15 minutes. For signal detection, samples were incubated in 10% blocking reagent (Roche, Mannheim, Germany) in PBS for 1 hour and afterwards for 30 minutes in blocking reagent containing anti-DIG alkaline phosphataseconjugated Fab fragment antibody (Roche) diluted to 1:500. After three washes for 10 minutes in PBS, tissues were equilibrated in detection buffer (100 mM Tris-HCl pH 9.5, 100 mM NaCl, 50 mM MgCl 2 ) for 15 minutes prior to incubating in the same buffer supplemented with 0.2 mM NBT and 0.2 mM BCIP substrates until a signal was visible.

Generation of plants overexpressing miR172
For miR172 overexpression, a 133 bp fragment of a putative miR172 precursor from potato was PCR-amplified using EST600546 (GenBank accession number BQ114970) as a template and primers miR172 forward 5Ј-GGTCTAGACATACAGTTGTTGCTTGCTA-3Ј and miR172 reverse 5Ј-GGGTCGACATCAAGTCATCAATTTGCCA-3Ј. The PCR product was digested with XbaI and SalI and cloned in sense orientation in the XbaI and SalI sites of pBIB35S-Hyg to generate pBIB-35S::miR172. The pBIB35S-Hyg vector was created by replacing the EcoRI/HindIII cassette of pBIB-HYG (Becker, 1990) with the EcoRI/HindIII cassette of pBINAR (Höfgen and Willmitzer, 1990). pBIB-35S::miR172 was electroporated into Agrobacterium tumefaciens strain C58 GV2260, which was used to transform S. tuberosum subspecies andigena essentially as described (Banerjee et al., 2006a). Hygromycin-B-resistant shoots were regenerated and then propagated from single-node stem cuttings on MS medium containing 20 g l -1 sucrose, 15 mg l -1 hygromycin B, 250 mg l -1 cefotaxime and 2 g l -1 Gelrite (Duchefa Biochemie). Plants were propagated for at least two rounds on selective medium before being propagated in the absence of the antibiotic. Twenty-three putative transgenic lines were planted in soil and the level of miR172 was assessed by RNA blot. Seven 35S::miR172 lines and one line carrying an empty vector were selected for further analysis.

Analysis of flowering and tuberization
To analyze flowering time, plants were grown either in the LD greenhouse or in controlled environment chambers with the following photoperiods: SD (8 hours light and 16 hours darkness); and SD supplemented with a 30 minutes white-light night break given 8 hours after the start of the dark period (SD+NB). Lighting was provided by high-pressure sodium vapor lamps (SON-T Agro 400 W). The shoot apex was carefully checked for

RESEARCH ARTICLE
Development 136 (17) visible signs of flowering every 2 or 3 days. Flowering time was measured as the number of days from planting in soil to the appearance of the floral bud. Leaf number was also recorded at the time the floral bud was visible.
For tuberization experiments, potato plants grown in the greenhouse were transferred to SD or SD+NB controlled environment chambers at the 10-12 leaf stage, approximately 4 weeks after potting. LD experiments were performed in the greenhouse. Tuberization was analyzed once per week under SD and SD+NB conditions and after 4 months of growth under LDs. Tuberization time was measured as the number of days from transfer to SD or SD+NB conditions to the appearance of tubers. Leaf number was also recorded when tubers were first visible.

Grafting experiments
Five-week-old LD-grown 35S::miR172 (line 8) and wild-type plants were used. V-shape grafts were made, bound with paper surgical tape, and immediately covered up with a transparent plastic bag to preserve humidity. Grafts were maintained in the LD greenhouse. Humidity was slowly released after 48 hours. Ten days after grafting, when graft unions had healed, the stock leaves were removed and plants were transferred to SD+NB conditions. Twenty grafts of each of the following types were analyzed: 35S::miR172 scions onto wild-type stocks, wild-type scions onto 35S::miR172 stocks, wild-type scions onto wild-type stocks and 35S::miR172 scions onto 35S::miR172 stocks. Ungrafted plants were also used as controls. Tuberization time was determined as described above.

Identification of RAP1, a putative miR172 target gene
In order to isolate potential target genes of miR172, total RNA was extracted from potato tissues with low miR172 levels. RNA was reverse transcribed using the RNA-ligase-mediated method from the Gene Racer Kit (Invitrogen). First, a forward primer corresponding to the miR172 sequence (5Ј-AGAATCTTGATGATGCTGCAT-3Ј) was used in a 3Ј RACE reaction to amplify the 3Ј part of cDNAs representing potential targets of miR172. A PCR product was gel-excised, cloned and sequenced. Then, a reverse primer (5Ј-GTCAAGAGTTGTCGAAGCAA-3Ј) designed on the obtained sequence was used for 5Ј RACE to obtain the full-length sequence of the cDNA. This sequence was named RAP1 and has been submitted to the EMBL Nucleotide Sequence Database under the accession number FM246879.

RESULTS
A miR172 precursor is present in potato A putative miR172 precursor, which contains a sequence identical to miR172a from Arabidopsis, had been previously identified in potato (Aukerman and Sakai, 2003). A prediction of the secondary structure of this putative precursor using Mfold showed that the miR172 sequence is present, paired to the miR172* sequence, in an arm of the stem of a stem-loop structure (see Fig. S1 in the supplementary material), as expected for a miRNA precursor. This suggests that miR172 can be produced in potato.

Analysis of miR172 levels in potato plants
The relative abundance of miR172 was analyzed by RT-qPCR and RNA blot, using a miR172 antisense probe, in potato plants grown under LD, SD+NB and SD conditions. As shown in Fig. 1 and Fig. S2 in the supplementary material, miR172 is present in all organs studied (leaves, stems and stolons), with a higher accumulation in stems under all photoperiods. miR172 levels were higher in all organs under tuberinducing conditions (SDs) than under non-inducing LDs (Fig. 1A), whereas under moderately inductive SD+NB conditions, levels were lower than under LDs (Fig. 1B). Both under SD and SD+NB conditions, an increase in miR172 levels correlated with tuber initiation, as its abundance was higher in swollen than unswollen stolons ( Fig. 1;  To examine the spatial distribution of miR172, we performed in situ hybridization on leaves, stems, stolons and swollen stolons. In leaves, miR172 was mainly detected in the epidermis and vascular cells ( Fig. 2A). By contrast, in the stem, a signal corresponding to miR172 was detected only in the internal and external phloem (Fig.  2B). In stolons collected under LDs, miR172 was present mainly in vascular cells and in the apical meristem (Fig. 2C). Under SDs, however, miR172 was detected in all cell types of unswollen stolons (Fig. 2D) and its localization was restricted to the vasculature and apex when stolons started to swell (Fig. 2E).

RESEARCH ARTICLE
Induction of tuberization by miR172 Altogether, these experiments show that: (1) miR172 is present in the vascular bundles of all the organs studied; (2) miR172 levels are highest under tuber-inducing SD conditions; and (3) miR172 distribution changes and its levels increase in stolons at the onset of tuberization.

Effect of PHYB on miR172 levels
The photoreceptor PHYB is involved in tuberization control in potato. To study whether PHYB regulates miR172 levels, we compared the accumulation of miR172 in different organs of wildtype and anti-PHYB plants under LD conditions by RNA blot and RT-qPCR. These experiments showed that the level of miR172 is reduced in the leaves of anti-PHYB plants (Fig. 3A). Moreover, miR172 abundance was dramatically higher in swollen than unswollen stolons of anti-PHYB plants (Fig. 3A). This difference was confirmed by RT-qPCR by comparing miR172 levels in stolons at 35 and 39 days in wild-type and anti-PHYB plants (Fig. 3B). The difference of miR172 abundance between swollen and unswollen stolons is more dramatic in anti-PHYB plants under LDs (Fig. 3) than in the wild type under SD or SD+NB conditions (Fig. 1). These experiments show that in plants in which PHYB is silenced, miR172 levels are reduced in leaves and increased in swollen stolons.

miR172 affects flowering time in potato
To understand the function of miR172 in potato, we generated plants that constitutively express miR172 from the cauliflower mosaic virus 35S promoter. Seven 35S::miR172 lines and a control line transformed with an empty vector were selected for analysis. The level of miR172 in the transgenic lines was examined by RNA blot and RT-qPCR. No change in the abundance of miR172 was observed in the control line compared to the wild type. By contrast, an increase in the amount of this miRNA was clearly visible in all the 35S::miR172 lines (see Fig. S3 in the supplementary material).
Since miR172 is a regulator of flowering time (Aukerman and Sakai, 2003;Chen, 2004;Lauter et al., 2005), we investigated the flowering time of potato 35S::miR172 plants. All the 35S::miR172 lines flowered earlier than control plants under LDs when flowering time was estimated by counting the number of leaves at flowering (Fig. 4A), and most lines were also statistically different from the controls in the number of days to flowering (Fig. 4B). Similar results were obtained under SD+NB and SD conditions (see Fig. S4 in the supplementary material). These results indicate that miR172 promotes flowering in potato, as previously shown in other plant species.

Overexpression of miR172 promotes tuberization in a photoperiod-dependent manner
Given that the regulation of flowering and tuberization shows several similarities (Suárez-López, 2005;Rodríguez-Falcón et al., 2006), we investigated whether the increase in miR172 levels could have an impact on tuber induction in potato. Under strongly inductive SD conditions, no significant differences in tuberization time were observed between 35S::miR172 lines and the controls (Fig. 5A,B). In moderately inductive SD+NB conditions, several 35S::miR172 lines (6, 8 and 22) showed an early tuberization

RESEARCH ARTICLE
Development 136 (17) LDs (C). Hybridization signal appears as purple coloration. am, apical meristem; c, cortex; e, epidermis; ep, external phloem; ip, internal phloem; lp, leaf primordium; pa, parenchyma; vc, vascular cells; x, xylem. Scale bars: 0.1 mm. phenotype (Fig. 5A,B). In addition, 35S::miR172 lines can tuberize in LD conditions, which are not inductive for the wild type; Fig. 5C shows the underground part of two 35S::miR172 lines in comparison to control plants after 3 months in LD conditions. Therefore, miR172 promotes tuber development and is probably involved in the photoperiodic control of tuberization. Although 35S::miR172 plants tuberize earlier than wild-type plants under SD+NB conditions and form tubers under LDs, they tuberize later under LD than SD+NB conditions (not shown) and later under SD+NB conditions than under SDs (Fig. 5A,B). Therefore, miR172 overexpression causes a partial loss of the photoperiodic control of tuber induction. Given that PHYB regulates miR172 levels (Fig. 3) and plays an important role in photoperiodic tuberization (Jackson et al., 1996), the tuberization time of anti-PHYB plants was determined under the three photoperiods. Anti-PHYB plants tuberized at the same time under SDs and LDs, indicating that they lost the repression of tuberization under LDs completely (see Fig. S5 in the supplementary material). Tuberization of anti-PHYB plants was very slightly delayed under SD+NB conditions compared with SDs or LDs (see Fig. S5 in the supplementary material). Therefore, the photoperiodic control of tuberization is more severely affected in anti-PHYB than 35S::miR172 plants.

The effect of miR172 on tuberization is graft transmissible
The presence of miR172 in vascular bundles (Fig. 2), the change in distribution in stolons at the onset of tuberization (Fig. 2) and the opposite effect of PHYB on miR172 levels in leaves and stolons (Fig. 3) suggested that this miRNA might be involved in longdistance signaling to control tuberization. Therefore, we tested whether the tuberization phenotype of 35S::miR172 plants is graft transmissible. 35S::miR172 scions grafted onto wild-type stocks (35S::miR172/WT grafts) tuberized as early as 35S::miR172/35S::miR172 controls (Fig. 5D), showing that the effect of miR172 on tuberization is transmissible through grafts and indicating that miR172 overexpression in aerial organs is sufficient to promote tuberization. By contrast, WT/35S::miR172 grafts tuberized like WT/WT controls (Fig. 5D), indicating that the overexpression of miR172 in underground parts is not sufficient to affect this process.
Identification of RAP1, a potato gene with a miR172 binding site All of the target genes of miR172 identified so far belong to the AP2-like family (Aukerman and Sakai, 2003;Schmid et al., 2003;Chen, 2004;Lauter et al., 2005;Chuck et al., 2007). In order to further investigate the role of miR172 in potato, we used an approach to identify potato genes possessing a miR172 binding site, which therefore represent candidate genes involved in the miR172 pathway. The miR172 sequence was used as a primer in a 3Ј RACE reaction, leading to the amplification of a cDNA fragment containing a miR172 binding site. The full-length cDNA sequence, as well as validation of the miR172 binding site, were obtained by subsequent 5Ј RACE. Using this approach, we identified a gene encoding a 454 amino acid protein showing strong homology with AP2 and AP2-like proteins. This protein was named RAP1 (RELATED TO APETALA2 1) and it contains two AP2 domains highly similar to those of previously characterized members of the AP2 family (Fig. 6A). The RAP1 coding sequence displays a miR172 binding site, and alignments with known targets of this miRNA indicated a partial conservation of mismatches between miR172 and its targets (Fig. 6B). This suggests that RAP1 is a likely target of miR172 in potato.

Transcript levels of RAP1 are regulated by miR172 and PHYB
To determine whether miR172 regulates RAP1, RT-qPCR analyses were performed. In wild-type plants grown under SD+NB or LD conditions, the mRNA abundance of RAP1 was higher in leaves than in stems and stolons (Fig. 7A). Therefore, RAP1 transcript levels show an inverse correlation with miR172 accumulation, suggesting that this miRNA can target RAP1 mRNA for degradation. The only significant difference in RAP1 mRNA abundance between SD+NB and LD conditions was observed in leaves, with higher levels under LD than SD+NB conditions (Fig. 7A).
To test whether RAP1 is a target of miR172, the levels of RAP1 transcript were analyzed in 35S::miR172 plants. RAP1 mRNA abundance was significantly lower in 35S::miR172 than in wild-type leaves (Fig. 7C), indicating that miR172 negatively regulates RAP1. By contrast, RAP1 mRNA levels were not significantly altered in the stems and stolons of 35S::miR172 plants (Fig. 7C).
Then, levels of RAP1 mRNA were examined in anti-PHYB plants, as these plants showed altered miR172 accumulation (Fig. 3). Levels of RAP1 were drastically reduced in all analyzed organs of anti-PHYB plants in contrast with the controls 2877 RESEARCH ARTICLE Induction of tuberization by miR172  ( Fig. 7D). Nevertheless, anti-PHYB plants still have lower levels of RAP1 in stems and stolons than in leaves (Fig. 7D). These results indicate that PHYB regulates RAP1.
Transcript levels of BEL5 are regulated by miR172 and PHYB Transcript levels of BEL5 were analyzed in different organs and photoperiodic conditions because BEL5 induces tuberization in potato (Chen et al., 2003;Banerjee et al., 2006b). RT-qPCR experiments showed that the levels of BEL5 mRNA were higher in leaves and stems than in unswollen stolons of wild-type plants under SD+NB and LD conditions (Fig. 7B), and an increase in BEL5 in stolons correlated with the onset of tuber development in SD+NB conditions (Fig. 7B; compare stolons with swollen stolons). This is consistent with a previous analysis of BEL5 transcript abundance (Chen et al., 2003). Wild-type leaves and stems showed a higher accumulation of BEL5 mRNA under LD than SD+NB conditions, whereas the opposite was found in stolons (Fig. 7B).
Since miR172, PHYB and BEL5 influence tuberization, we examined whether miR172 and PHYB have an effect on BEL5. Under SD+NB conditions, the levels of BEL5 mRNA were higher in all organs of 35S::miR172 than wild-type plants (Fig. 7E). Therefore, miR172 overexpression leads to the upregulation of BEL5. In addition, in anti-PHYB lines the levels of BEL5 were increased in stems and stolons when compared with the wild type, whereas in leaves transcript levels were decreased in one of the transgenic lines, anti-PHYB-10 (Fig. 7F). In other experiments, a reduction in BEL5 mRNA was also observed in the leaves of transgenic line anti-PHYB-4. Interestingly, there was a remarkable increase in BEL5 levels in anti-PHYB stolons at the onset of tuberization (Fig. 7F, compare stolons with swollen stolons). These results indicate that PHYB controls BEL5 transcript abundance in several organs. The reduction in BEL5 mRNA levels in the leaves of anti-PHYB plants and the increase in mRNA levels in stolons when these start to swell is consistent with the reported movement of BEL5 mRNA (Banerjee et al., 2006b).

DISCUSSION
It has been reported that miR172 regulates flowering time, flower development, vegetative phase change, sex determination and meristem cell fate (Aukerman and Sakai, 2003;Chen, 2004;Lauter et al., 2005;Chuck et al., 2007;Zhao et al., 2007). We show here that, in potato, miR172 affects not only flowering time, but also the timing of tuberization, another developmental process related to reproduction.
We have overexpressed a member of the miR172 family that had been previously identified in potato (Aukerman and Sakai, 2003). Recently, this miR172 member has been named miR172b, and four additional potato miR172 precursors have been computationally

RESEARCH ARTICLE
Development 136 (17) identified (Zhang et al., 2009). The predicted secondary structure of pre-miR172b shown in Fig. S1 in the supplementary material is identical to that reported by Zhang et al. (Zhang et al., 2009). The sequence of potato miR172b is identical to that of Arabidopsis miR172a and b and maize miR172a-d. However, overexpression of this miRNA in potato does not cause any obvious floral organ identity phenotype (not shown), suggesting that the function of this miRNA has partially diverged between potato and other species.
Our results show that overexpression of this miRNA promotes tuberization and suggest that it impairs the response of tuberization to the photoperiod, as 35S::miR172 plants tuberize under non-inductive LDs and tuberize earlier than the wild type under SD+NB conditions, but not under SDs (Fig. 5). Consistent with the promotion of tuberization by miR172, the levels of this miRNA are higher under strongly inducing SDs than under LDs, and an increase in its levels in stolons correlates with early stages of tuber development under SD and SD+NB conditions in wildtype plants (Fig. 1). In addition, tuberization of anti-PHYB plants under LDs is associated with high miR172 abundance in stolons when these start to swell (Fig. 3). In Arabidopsis, miR172 levels are increased in a phyB mutant (Jung et al., 2007), indicating that the regulation of miR172 by PHYB is conserved in different plant species.
miR172 promotes flowering in Arabidopsis by negatively regulating AP2-like flowering repressors, such as TOE1, TOE2, SMZ and SNZ (Aukerman and Sakai, 2003;Schmid et al., 2003;Jung et al., 2007). Since this miRNA accelerates flowering and tuberization in potato (Figs 4 and 5), we propose that miR172 would negatively regulate flowering and tuberization inhibitors in this species. We have cloned a potato AP2-like gene, RAP1, that contains a miR172 target site. There is an inverse correlation between the abundance of RAP1 transcript and miR172 in several organs of wild-type plants ( Fig. 1; Fig. 7A). Moreover, downregulation of RAP1 in miR172-overexpressing leaves (Fig.  7C) strongly suggests that RAP1 is a target of miR172 and that this miRNA induces the degradation of RAP1 mRNA. The finding that RAP1 transcript abundance is not reduced in 35S::miR172 stems and stolons (Fig. 7C), which do overexpress miR172 (see Fig. S3B in the supplementary material), does not exclude the possibility that miR172 can regulate RAP1 in these organs by blocking its translation. In fact, miR172 can induce both mRNA degradation and translational inhibition in other species (Aukerman and Sakai, 2003;Kasschau et al., 2003;Chen, 2004;Lauter et al., 2005;Schwab et al., 2005;Chuck et al., 2007). It is also possible that feedback regulation of miR172 targets by their products masks changes in target mRNA abundance, as has been shown in Arabidopsis (Schwab et al., 2005).
Silencing of PHYB completely abolishes the repression of tuberization under LDs (Jackson et al., 1996) (see Fig. S5 in the supplementary material), indicating that PHYB is essential for the inhibition of tuberization under this photoperiod. miR172 levels are altered in anti-PHYB plants (Fig. 3), strongly suggesting that miR172 acts downstream of PHYB to control tuberization. However, changes in miR172 levels are not sufficient to explain the strong effect of PHYB on tuberization, as the tuberization phenotypes of 35S::miR172 plants are weaker than those of anti-PHYB plants and, in addition, 35S::miR172 plants are still partially sensitive to day length (Fig. 5;see Fig. S5 in the supplementary material). Similarly, Arabidopsis plants overexpressing miR172 also show a residual photoperiodic flowering response (Aukerman and Sakai, 2003). Given that BEL5 promotes tuberization (Chen et al., 2003;Banerjee et al., 2006b), we examined the possibility that miR172 and PHYB could regulate this gene. We found increased BEL5 mRNA levels in all organs of 35S::miR172 plants and in stems and stolons of anti-PHYB plants (Fig. 7E,F). Therefore, BEL5 might also contribute to the regulation of tuber induction by miR172 and PHYB. Taken together, these results suggest that the regulation of BEL5 by PHYB is at least in part mediated by miR172. Our results do not rule out that additional genes might act downstream of PHYB and miR172 to mediate their effects on tuber formation.
The presence of miR172 in the vasculature of wild-type potato plants (Fig. 2), the detection of miR172 in the phloem sap of Brassica napus  and the graft transmission of the tuberization phenotype of 35S::miR172 plants (Fig. 5D) show that either this miRNA is mobile or it regulates long-distance signals to induce tuberization. Our grafting experiments, together with the analysis of RAP1 and BEL5 mRNA levels in miR172overexpressing plants, indicate that miR172 can act at least in the leaves to promote tuberization and regulate the mRNA levels of these two genes. Although overexpression of miR172 in the stocks of WT/35S::miR172 grafts does not promote tuberization (Fig. 5D), this does not rule out the possibility that endogenous miR172 can act in these organs. The clear correlation of tuber induction with an increase in miR172 levels in stolons is consistent with this idea. It is possible that the overexpression of miR172 in stocks is not sufficient to counteract repressive signals coming from the wild-type scions.
It has been established that the FT protein is a component of the long-range signals for flowering (Turck et al., 2008). However, its counterpart for tuberization has not been identified so far. It has been reported that BEL5 mRNA is graft transmissible and its movement correlates with tuber induction (Banerjee et al., 2006b). Interestingly, anti-PHYB plants show a reduction in BEL5 transcript levels in leaves in comparison with wild-type plants and an increase in the levels in stolons when these start to swell (Fig. 7F). Similarly, we have observed reduced miR172 levels in leaves and increased levels in swollen stolons of anti-PHYB plants (Fig. 3). One interpretation of these results is that PHYB promotes miR172 and BEL5 expression in leaves and represses their expression in stolons. However, this is inconsistent with the fact that PHYB represses tuberization, whereas overexpression of miR172 and BEL5 in aerial organs promotes it. Therefore, we hypothesize that BEL5 mRNA and miR172 move from leaves to stolons under SDs to induce tuberization. Under LD and SD+NB conditions, this movement would be repressed by PHYB, thereby preventing or delaying tuberization. Previously it had been proposed that PHYB induces the production of a graft-transmissible inhibitor of tuberization (Jackson et al., 1998). Our data suggest that PHYB might inhibit the transport of promoters of tuberization. Although the idea that miRNAs can act as systemic signals is still under debate (Voinnet, 2009), at least one miRNA has been shown to be a long-distance signal (Pant et al., 2008).
Our results allow us to propose a model for the regulation of tuberization. miR172 is regulated by PHYB and these both regulate RAP1 and BEL5. This strongly suggests that PHYB acts upstream of miR172 and BEL5, and BEL5 acts at least in part downstream of miR172, to control tuberization. As suggested above, PHYB might repress the movement of BEL5 mRNA and miR172 from leaves to stolons under LDs. Under SDs, this repression would be released to allow BEL5 and miR172 to induce tuberization. This is consistent with the notion that PHYB acts in LD conditions to repress a tuberinducing pathway that is active under SDs (Jackson et al., 1996). miR172 promotes tuberization probably via the upregulation of BEL5 in leaves and stolons. Moreover, this miRNA inhibits RAP1. We postulate that the upregulation of BEL5 by miR172 might be mediated by a miR172 target gene, perhaps RAP1, which would repress BEL5. Interestingly, the BEL5 promoter contains two CAACA sequence motifs (in reverse orientation, TGTTG) that have been described as binding sites for the AP2 domain of an AP2-like transcription factor (Kagaya et al., 1999), and RAP1 has two AP2 domains. Increased miR172 levels lead to the inhibition of its targets (e.g. RAP1) and therefore probably to a release of BEL5 repression. This release would lead to more BEL5 mRNA in stolons, therefore inducing tuberization. However, we cannot rule out an effect of miR172 target genes on tuberization through a BEL5-independent pathway. In addition, PHYB regulates RAP1 probably through miR172, but also through a miR172-independent pathway, as the effect of PHYB on RAP1 mRNA levels is much more dramatic than that of miR172 (Fig. 7). This complex network probably confers robustness to the regulation of tuberization by reinforcing signals, as PHYB acts under LDs to repress the function of two tuberization inducers, BEL5 and miR172, and to promote the expression of a putative negative regulator of BEL5 (RAP1), ensuring that tuberization does not occur under LD conditions. This work establishes that miR172 is another regulator shared by flowering and tuberization, shows that this miRNA affects a long-distance signaling pathway and strongly suggests that it acts between PHYB and BEL5 to control tuber induction. In Arabidopsis, several miR172 target genes are involved in the

RESEARCH ARTICLE
Development 136 (17) regulation of flowering time. Further research will help to determine whether different miR172 targets are specialized for the control of flowering and tuberization.